Class ClientSession

tensorflow/cc/client/client_session.h:53–143

A `ClientSession` object lets the caller drive the evaluation of the TensorFlow graph constructed with the C++ API. Example: Scope root = Scope::NewRootScope(); auto a = Placeholder(root, DT_INT32); auto c = Add(root, a, {41}); ClientSession session(root); std::vector outputs; Status s = session.Run({ {a, {1}} }, {c}, &outputs); if (!s.ok()) { ... }

///     Status s = session.Run({ {a, {1}} }, {c}, &outputs);
///     if (!s.ok()) { ... }
class ClientSession {
 public:
  /// A data type to represent feeds to a Run call.
  ///
  /// This is a map of `Output` objects returned by op-constructors to the value
  /// to feed them with. See `Input::Initializer` for details on what can be
  /// used as feed values.
  typedef std::unordered_map<Output, Input::Initializer, OutputHash> FeedType;

  /// Create a new session to evaluate the graph contained in `scope` by
  /// connecting to the TensorFlow runtime specified by `target`.
  ClientSession(const Scope& scope, const string& target);

  /// Same as above, but use the empty string ("") as the target specification.
  ClientSession(const Scope& scope);

  /// Create a new session, configuring it with `session_options`.
  ClientSession(const Scope& scope, const SessionOptions& session_options);

  ~ClientSession();

  /// Evaluate the tensors in `fetch_outputs`. The values are returned as
  /// `Tensor` objects in `outputs`. The number and order of `outputs` will
  /// match `fetch_outputs`.
  Status Run(const std::vector<Output>& fetch_outputs,
             std::vector<Tensor>* outputs) const;

  /// Same as above, but use the mapping in `inputs` as feeds.
  Status Run(const FeedType& inputs, const std::vector<Output>& fetch_outputs,
             std::vector<Tensor>* outputs) const;

  /// Same as above. Additionally runs the operations ins `run_outputs`.
  Status Run(const FeedType& inputs, const std::vector<Output>& fetch_outputs,
             const std::vector<Operation>& run_outputs,
             std::vector<Tensor>* outputs) const;

  /// Use `run_options` to turn on performance profiling. `run_metadata`, if not
  /// null, is filled in with the profiling results.
  Status Run(const RunOptions& run_options, const FeedType& inputs,
             const std::vector<Output>& fetch_outputs,
             const std::vector<Operation>& run_outputs,
             std::vector<Tensor>* outputs, RunMetadata* run_metadata) const;

  /// \brief A handle to a subgraph, created with
  /// `ClientSession::MakeCallable()`.
  typedef int64 CallableHandle;

  /// \brief Creates a `handle` for invoking the subgraph defined by
  /// `callable_options`.
  /// NOTE: This API is still experimental and may change.
  Status MakeCallable(const CallableOptions& callable_options,
                      CallableHandle* out_handle);

  /// \brief Invokes the subgraph named by `handle` with the given options and
  /// input tensors.
  ///
  /// The order of tensors in `feed_tensors` must match the order of names in
  /// `CallableOptions::feed()` and the order of tensors in `fetch_tensors` will